When do a handful of anecdotal events observed in a complex system prove millions of unseen events? Almost never. Our vast time-tested literature on statistics must guide our extrapolations in these situations and anyone suggesting otherwise should be ignored. These were my thoughts as I read this Tweet by psychologist Scott Barry Kaufman. In a comment to this Tweet, Scott mentions these specific problem areas: “Tail ratios, normal distribution, statistical significance, probability, hypothesis testing . . .”
[1]
Print this post [2]
[3]